Before I worked in AI, I spent years in service design. Mapping customer journeys. Finding friction points. Understanding how work actually flows through organisations.
That background matters more than any technical AI knowledge I’ve picked up since.
AI implementation is really just service design with a new toolkit. You’re still solving for human experience. The technology is secondary.
The Pattern Nobody Talks About
I’ve watched internet adoption, mobile, cloud, and now AI roll through organisations. The pattern repeats every time:
The technology changes. The human capabilities that determine success stay the same.
Communication. Collaboration. Feedback. Understanding context. The ability to research and learn.
These connector skills matter more than the platform you choose or the model you subscribe to.
What I Actually Look For
When an SME brings me in because their AI implementation isn’t working, I don’t start with the technology.
I look for patterns of communication and preparation.
How are teams talking to each other? Are there silos? How do they share information? What channels exist for feedback?
You need to understand how people work before you change how they work.
If you’ve got everyone in silos and they start using AI, they’re still just using AI in silos. The tool doesn’t fix organisational dysfunction.
The Question Everyone Asks
People often ask me: which AI tool should we use? Claude or ChatGPT or Gemini or Perplexity or DeepSeek?
It really doesn’t matter.
You can do most things with most models. The tool is a small percentage of success when you’re doing standard work.
For specialised use cases, specific tools might work better. But for 80–90% of what SMEs need, the platform choice won’t determine your results.
Get the process right. Organise the people. The rest falls into place.
The Real Resistance
A common pattern: an SME leader worries that people are resisting AI. So they mandate use of AI tools.
But they’re not giving direction or vision. They’re just telling people to do stuff quicker.
They’re worried about adoption. They should be worried about coordination.
I’ve seen teams with Microsoft Copilot rolled out to everyone. Access sorted. Subscriptions paid. Nobody using it.
The problem isn’t training. When you start using these tools, you figure them out fairly quickly.
The problem is lack of shared vision about what you’re trying to achieve.
You can’t just throw tools at people and expect results.
What Service Design Actually Brings
Here’s what I do differently because of my service design background:
I start with the customer. Not the platform. Not the technology stack.
I map the journey. I look for the easiest route to achieving what the user wants to achieve. I focus on reducing friction in that journey.
This is basic UX thinking applied to AI implementation.
Most tech consultants optimise the platform. I optimise the experience.
Why SMEs Actually Have An Advantage
The industry narrative says SMEs are at a disadvantage with technology adoption. Too small. Not enough resources. Can’t move as fast as enterprises.
That’s backwards.
SMEs can pivot and deliver business change much faster than large organisations.
When you’ve got a dozen people in a team using a CRM or a particular system, you can swap it out quickly. You can talk to everyone involved. You can make better choices faster.
Business change is much faster in SMEs because you’re not navigating layers of approval and stakeholder management.
What Actually Works
The AI implementations that succeed have a few things in common:
Clear direction from leadership about what you’re trying to improve and why it matters.
Space for people to experiment without mandating specific outputs or timelines immediately.
Regular feedback loops where teams share what’s working and what isn’t.
Focus on coordination rather than just individual productivity gains.
The technology handles itself once you’ve sorted the organisational side.
That’s the bit most people get wrong. They start with the tool. They should start with the people and the process.
Service design taught me that years ago. AI hasn’t changed it.
